How to find the longest string in an array of variables?

[Vincenz Jahn]

Hi all,

I am looking for a command to detect the longest string or numeric variable in an array, and to generate a new variable with this string.

Thx for your help

[Aaditya Dar]

If you are referring to a column array then you could use the following procedure (-carryforward- is available from SSC).

gen length_var = length(var)
gsort -length_var
sum length_var
gen longest_var = var if length_var == `r(max)'
carryforward longest_var, gen(newvar)

[Nick Cox]

How do you decide whether a string or numeric variable is longer? Are you presuming a decimal representation and if so, how do you handle fractional parts?

[Vincenz Jahn]

With "longer" I mean the number of characters a variable has.

[Nick Cox]

Aaditya: Your interpretation of the problem lies in a different direction from mine. You seem to be assuming that the question is about the longest value in a string variable. I am assuming "array" is an unStataish description of a set of variables, some numeric, some string, but am still waiting for clarification of how to think of the length of a numeric variable.

That aside, your code will not work for numeric variables as soon as you apply length().

For string variables alone, you can get there with

  
gen length_var = length(var)
sort length_var
gen longest_var = var[_N]

with some modification no doubt for strLs.

[Nick Cox]

Vincenz #4: Sorry, but that doesn't help me. What is a character? Do you mean a byte?

[Aaditya Dar]

gen longest_var = var[_N]

Very neat, thanks!

[Vincenz Jahn]

Thank you for your help so far. I think I wasn't able to communicate the actual problem I have. Let's try it another time:
I have a set pretty similar to the following:

Var1	Var2	Var3	Var4
Ref. 54322	Ref. 5432	Ref. 543
Ref. AB562	Ref. AB56	Ref. AB5
	Ref. 1234	Ref.123	Ref.12

I need to generate a new variable that shows all the variables with the longest content, in terms of characters:

VarNew

Ref. 54322

Ref. AB562

Ref. 1234

Thank you for your help!

[Vincenz Jahn]

Character as either a letter, a number or other signs.

[Andrew Musau]

Have a look at the FAQs on how to use dataex to present data examples. Employing Nick's code in #5

input strL(var1 var2 var3 var4)
"Ref.54322" "Ref.5432" "Ref.543" ""
"Ref.AB562" "Ref.AB56" "Ref.AB5" ""
"" "Ref.1234" "Ref.123" "Ref.12"
end

gen id=_n
reshape long var, i(id)
by id: gen length_var = length(var)
sort id length_var
by id: gen longest_var = var[_N]
drop length_var
reshape wide var, i(id) j(_j)
l

. l

     +----------------------------------------------------------+
     | id        var1       var2      var3     var4   longest~r |
     |----------------------------------------------------------|
  1. |  1   Ref.54322   Ref.5432   Ref.543            Ref.54322 |
  2. |  2   Ref.AB562   Ref.AB56   Ref.AB5            Ref.AB562 |
  3. |  3               Ref.1234   Ref.123   Ref.12    Ref.1234 |
     +----------------------------------------------------------+

[William Lisowski]

Post #8 also posted at http://www.statalist.org/forums/foru...n-one-variable with a different answer given.